Understanding B-type supergiants in the low metallicity environment of the SMC II

0.44 MB·English

by  C. Trundle

Astronomy&Astrophysicsmanuscriptno.AA2061 February2,2008
(DOI:willbeinsertedbyhandlater)
Understanding B-type supergiants in the low metallicity
environment of the SMC II.
C.Trundle1,2 &D.J.Lennon1
1 TheIsaacNewtonGroupofTelescopes,ApartadodeCorreos321,E-38700,SantaCruzdeLaPalma,CanaryIslands,Spain
5 2 TheDepartmentofPureandAppliedPhysics,TheQueen’sUniversityofBelfast,BelfastBT71NN,NorthernIreland
0
0
2 Abstract.Despitearesurgence of effort over thelastdecade intheareaof massivestarsthereisstillambiguityover their

n evolutionary path,contamination of theirsurfaceabundances and thebehaviour oftheirstellarwinds. Here10SMCB-type
a supergiantsareanalysedapplyingaunifiedmodelatmospherecodeFASTWINDtointermediateresolutionspectrafromtheESO
J MultiModeInstrument(EMMI)ontheNTTtelescope.Combinedwiththe8targetsanalysedinpaper1(Trundleetal.2004),

2 thiswork provides observational results on theproperties of the winds and chemical compositions of B-type supergiants in
1 theSMC.Thispaperemphasizesandsubstantiatestheimplicationsforstellarevolutionfrompaper1;thatcurrenttheoretical
modelsneedtoproducelargerdegreesofsurfacenitrogenenhancementsatlowerrotationalvelocities.Inadditionasignificant

1 discrepancy between theoretical and observed mass-loss rates is discussed which will have important implications for the
v
rotational velocities obtained from stellar evolution calculations. Furthermore, an initial calibration of the wind-momentum
8
luminosityrelationshipforB-typesupergiantsinalowmetallicityenvironment(Z=0.004)ispresented.
2
2
1
0 Keywords.stars:atmospheres–stars:early-type–stars:su- Kudritzki, Lennon& Puls (1995).This work showed that the
5 pergiants–stars:mass-loss–stars:abundances–stars:evolu- wind-momentum of a star, a product of its mass-loss, termi-
0
tion nalvelocityandradius,isdirectlyrelatedtoitsluminosityand
/
h can therefore provide an estimate of the stellar distance. The
p intrinsicallyhighluminosityofearly-typesupergiantscoupled
- 1. Introduction
o withtheirrelativelynormalspectralbehaviourmakesthemthe
r Theevolutionofagalaxyisstronglyinfluencedbytheinputof best candidates for using this Wind-momentum Luminosity
t
s energy,momentumandprocessedmaterialfrommassivestars Relationship (WLR). Recentworkhas shownthe dependence
a
oftheWLRonspectraltypeforGalacticOBA-typestars(Puls
: intotheinterstellarmedium(ISM).Aspossibleprogenitorsof
v et al. 1996; Kudritzki et al. 1999; Repolust, Puls & Herrero
TypeIIsupernovaeandgammaraybursts,thenucleosynthesis
Xi processesandevolutionofmassivestarsareimportanttopicsin 2004; Markova et al. 2004). Some studies have also endeav-
oured to understand the behaviour of the WLR in the lower
r currentastrophysicalresearch.Inspiteofitsimportance,there
a metallicityenvironmentsoftheMagellanicClouds(Pulsetal.
is still a large degree of ambiguity concerning the evolution-
1996; Kudritzki & Puls 2000; Crowther et al. 2002; Trundle
ary path of these objects beyond the main-sequence and the
etal.2004;Evansetal.2004a).Mostofthesestudiesfocused
influenceofrotationandmass-lossonthecompositionoftheir
on relativelysmall samplesof O-type stars with some indica-
stellar photospheres. Hence quantitative observationalstudies
tion of the WLR having different behaviour at lower metal-
concerning the chemical compositions and winds of massive
licity. Recent work by Trundle et al. (2004; hereafter paper
stars are vital tests on both the theory of galactic and stellar
1) investigated a sample of B-type supergiants in the SMC.
evolution. These early-type stars are also important contribu-
Althoughthesampleonlyincluded8targets,itprovidedobser-
tors to the spectra of star-forming galaxies and intermediate
vationalevidencetotestlinedrivenwindtheoryinthisspectral
age (20-50 Myr) stellar populations (de Mello et al. 2000).
region and at a low metallicity. Indeed this work highlighted
Furthermoreinordertomodeltheintegratedlightofstarburst
asignificantdiscrepancybetweentheobservedmass-lossrates
regions and constrain the composition of high redshift galax-
and the theoretical predictions by Vink et al. (2001) at SMC
ies,accurateatmosphericandwindparametersofOBstarsare
metallicities.
required(Leithereretal.1999;Pettinietal.2000).
In the mid 1990’s an intriguing implication arose from
Inadditiontowindanalyses,Paper1investigatedthechem-
the theory of radiatively driven winds and was introduced by
ical compositionof the SMC sample. This providedquantita-
Sendoffprintrequeststo:[email protected]. tive CNO abundancesof B-type supergiants;a key test of the
RecentlymovedtoInstitutodeAstrofis´icadeCanarias,C/V´iaLa´ctea, mixing and mass-loss processes incorporatedin stellar evolu-
E-38200LaLagunaTenerife,Spain. tion modelsfor early-typestars. The results from Paper 1 are
2 C.Trundle&D.J.Lennon:SMCB-typeSupergiantsII
Table1. Observationaldetails.IdentificationnumbersarefromAzzopardi&Vigneau(1982;AV#)andSanduleak(1968;Sk#).
Spectral types adopted from Lennon (1997), except for AV78 which has been reclassified in this work from the blue spectra.
AbsoluteMagnitudes(Mv)arecalculatedusingVand(B-V)magnitudesfromGarmany,Conti&Massey(1987;1)andMassey
(2002;2) and (B-V) values from Fitzpatrick & Garmany (1990). The adopted distance modulus is 18.9 (Harries, Hilditch, &
0
Howarth 2003). The average S/N ratios for the blue and red arms are presented. The vsini values representthe width of the
spectralfeaturesandaresimplyupperlimitsontheprojectedrotationalvelocities(seePaper1).
STAR Alias Spectral V B-V Mv S/N vlsr vsini
Type B R (kms−1) (kms−1)
AV420 Sk131 B0.5Ia 13.092 -0.17 -5.90 84 49 188±12 80
AV242 Sk85 B1Ia 12.111 -0.13 -6.91 125 47 188±25 90
AV264 Sk94 B1Ia 12.361 -0.15 -6.60 102 74 142±14 85
AV78 Sk40 B1Ia+ 11.051 -0.03 -8.25 161 80 176±12 85
AV96 Sk46 B1.5Ia 12.592 -0.10 -6.50 146 69 161±10 90
AV373 Sk119 B2Ia 12.171 -0.09 -6.92 80 105 191±26 80
AV10 Sk7 B2.5Ia 12.582 -0.02 -6.69 80 110 225±21 85
AV56 Sk31 B2.5Ia 11.151 -0.00 -8.18 120 100 209±11 80
AV443 Sk137 B2.5Ia 10.971 -0.06 -8.18 165 160 191±22 73
AV151 Sk57 B2.5Ia 12.262 -0.02 -7.01 122 80 187±15 62
consistent with the significant enhancements and dispersions per pixelin the blue and red arms, respectively(see Table 1).
of nitrogen abundances, previously observed in B-type stars Further details of the observations and data reduction proce-
(Gies&Lambert1992;Lennonetal.1991,1996,1997,2003; durescanbefoundinLennon(1997).
Fitzpatrick & Bohannan 1993; McErlean, Lennon, & Dufton TheoriginaldataselectedbyLennon(1997)areasubsetof
1999;Duftonetal. 2000).Theexistenceofa stellar windcan B0 - B9 starsfromthe sampleofGarmanyetal.(1987),who
reducetheangularmomentumofastar,whichinturncausesa carriedoutaspectralclassificationofSMCOB-typestars.For
decreaseintherotationalvelocity.Thislinkbetweenthemass- thepurposeofthispapertenstarsfromthisdatasethavebeen
lossof a star and its rotationalvelocityaffectsthe stellar life- selected spanningthe spectralrangeB0 - B2.5. An additional
timesandphotosphericcompositionpredictedbystellarevolu- constraint was that each object have medium to strong stellar
tioncodes.Thusmass-lossrates,rotationalvelocitiesandabun- winds,whichwasapparentfromtheprofileoftheHα line.In
dancesareimportantobservationalconstraintsforsuchcodes. Table1thetargetsarelistedwiththespectraltypesassignedby
The recent inclusion of rotationally induced mixing in stellar Lennon(1997)fromapplyinghisrevisedclassificationsforB-
evolution codes such as those by Maeder & Meynet (2001), typesupergiantsintheSMC. AV78hasbeenreclassifiedhere
wereshowninpaper1toreproducethelargedispersionofni- asaB1Ia+ starduetothepresenceofaweakSiIV lineinits
trogen abundances observed in B-type supergiants, implying spectrumat4116A˚.ThisstarwaspreviouslyclassedasaB1.5
thatrotationmayplayasignificantroleintheevolutionofmas- Ia+ starbyLennon(1997).
sivestars.
Theestimatedprojectedrotationalvelocities(vsini)were
The dataset of Lennon (1997) offers the opportunity of
determinedusing the techniquesdescribed in paper 1 and are
analysing an additional ten B-type supergiants which supple-
presented in Table 1. It is important to note that the vsini
mentstheresultsfrompaper1andwillprovideaclearerinsight
values only represent the width of the spectral features and
intothebehaviouroftheseluminousobjects.Theobjectiveof
are simply upper limits for the projected rotational velocities
this work is to provideobservationalconstraintson the mass-
(i.e. no attempt has been made to deconvolve the contribu-
lossratesofB-typesupergiantsandtocalibratetheWLRatthe
tion of stellar rotation and macroturbulent motions from the
metallicityoftheSMC.
line broadening).Equivalent widths were measured by fitting
a Gaussian profile using a non-linear least squares technique
in DIPSO (Howarth et al. 2003);a spectrum analysis package
2. ObservationalData
fromSTARLINK.Table1alsoliststheabsolutemagnitudescal-
The SMC optical dataset consideredfor this work is a subset culated using V and (B-V) magnitudes from Garmany, Conti
ofthatpresentedbyLennon(1997),inastudytodelineatethe &Massey(1987)andMassey(2002)and(B-V)0 valuesfrom
spectral classifications of B-type supergiants in low metallic- Fitzpatrick&Garmany(1990).Theadopteddistancemodulus
ityenvironments.Themediumresolutionspectra(R∼20000) oftheSMC,appliedtodeterminetheabsolutemagnitude(Mv),
were obtained remotely on the NTT telescope with the ESO is18.9(Harries,Hilditch,&Howarth2003).
Multi Mode Instrument (EMMI). Three spectral regions were Radialvelocitiesofthetargetstarsarecorrectedtothelocal
observed covering the wavelengths 3925-4375 A˚, 6190-6830 standardofrest,givingarangeinvelocitiesof142-225kms−1
A˚ and 4300-4750A˚. Signal-to-noise(S/N) ratiosgreater than (notesimilarrangeobservedforSMCstarsinPaper1).Froma
70 were obtained along with dispersions of 0.45 and 0.32 A˚ H I 21cmemissionsurveybyMcGee&Newton(1981),four
C.Trundle&D.J.Lennon:SMCB-typeSupergiantsII 3
gas complexeswere identified in the SMC at heliocentric ve- of ξ and thereforea value of 10 kms−1 was adoptedin the
O
locitiesof114,134,167&192kms−1.InadditionWeltyetal. analysisforthisstar.
(1997) found SMC multiple absorption components of inter- Threetargets,AV242,AV264&AV78,wereincludedinthe
stellargaswithvelocitiesbetween85≤and210kms−1.This UV analysisofEvansetal. (2004- hereafterEVANS04)(de-
impliesthattheradialvelocitiesofourtargetstarsareconsis- scribed in Paper 1), using archive IUE/HIRES data for AV78
tentwiththoseoffieldstarsintheSMC. and HST/GHRS spectra for the former two stars. These ob-
jects, therefore have predetermined terminal velocities mea-
sured with the SEI (Sobolev Exact Integration) method from
3. Stellar parameters
the UV resonance lines. The remaining seven stars have no
The model atmospheres and procedures used in this analysis availableUVdatafromwhichterminalvelocitiescanbemea-
areidenticaltothosedescribedinPaper1andwillonlybriefly suredandhenceestimatesweremadeusingthealternativepro-
be repeated here. The ‘unified model atmosphere’ code ap- cedure described in Paper 1 & EVANS04. This involves cal-
plied is called FASTWIND, a line-blanketed, spherically sym- culatingtheescapevelocityandapplyingtheratiosofv∞/vesc
metric,non-LTEcode(Santolaya-Rey,Puls,&Herrero(1997); determinedbyKudritzki& Puls(2000)(viz.forT ≥ 21kK
eff
Herrero, Puls, & Najarro 2002; Repolust, Puls, & Herrero thisratioisequalto2.65andforT <21kKitis1.4).
eff
2004). The temperature structure in FASTWIND was not cal- Hα profilesprovideameanstodeterminethepropertiesof
culatedexplicitly,butparameterisedbyapplyingthenon-LTE thestellarwind;withaknowledgeoftheterminalvelocity,the
Hopf function described by Santolaya-Rey, Puls, & Herrero mass-loss rate (M˙) and beta parameter (β) can be determined
(1997; see equations 4.1 - 4.3) to TLUSTY models calculated simultaneously from the profile shape. M˙ controlsthe overall
at Queen’s University Belfast (Ryans, R.S.I.; private commu- windemissionintheprofile,whilstβ causesavariationinthe
nication). strengthandFWHMoftheemissionpeak.Assuchitisdifficult
The stellar parameterswere calculatedusing preciselythe toconstrainthebetaparameterinstarswithweakwinds,where
samemethodsasdescribedinPaper1.Inshort,theparameter windemissionmayonlycause asmallamountof fillingin of
spacein whicheachstarresideswasestimatedusinga coarse theabsorptioncore(viz.AV420;seeFig1).Infactavariation
gridandthiswasfollowedbyaniterativeprocesstodetermine of β from 0.8 to 2 for AV420 with subsequentchanges in M˙
each parameter accurately in a finer but smaller grid. The re- from0.49to0.14.10−6M yr−1 willreproducetheobserved
⊙
sulting photospheric and wind parameters for each target are Hα profile. In this analysis we adopt a β parameter of 1 for
presentedinTable2. this weak wind star with the knowledge that this will lead to
FortheB0.5-B2stars,effectivetemperaturesweredeter- uncertainmass-lossrates(seealsoAV216andAV104inPaper
minedfromtheSiIII/SiIVionisationbalancewithtypicalran- 1).
dom errors of 1.5 kK. In the B2 - B2.5 stars only the Si III FortheP-Cygniandemissionprofiles,βcanbedetermined
lineswerepresent,asa resultprofilefitting tothese lineswas rather more accurately on the order of ± 0.5. This is consid-
required to determine the temperature. For these objects, the erably higher than the error found from the UVES data and
lackofaSi IV 4116A˚ lineandtheveryweakfeaturesforthe is attributed to the lower resolution and in some cases lower
SiII4128,4131A˚ profilesactedasupperandlowerlimitsfor S/N of the current dataset. In addition, nebular contributions
thetemperatureestimates.TheonlyexceptiontothisisAV151 in AV96& AV264preventaccurateestimatesof the emission
whichexhibitedrelativelystrongSiII featuresallowingtheSi peakFWHMandaβ intherange2-3cangenerallyfitthese
II/Si III ionisation balance to be determined. The uncertainty profileswithvaluesabove3havinglittleimpactontheprofile
for AV151 is 1.5 kK, and is slightly larger for the other B2- shape.SinceM˙ andβ arederivedsimultaneouslyfromthefit-
B2.5 stars (2.0 kK). This uncertainty in effective temperature tingofHα theerrorinthemass-lossrateisdominatedbythis
alsoaffectsthedeterminationofthelogarithmicsurfacegrav- uncertaintyinβ.ThedifferenceinM˙ derivedwithaβof2and
ity,typicallyanerrorof0.5kKrepresentsasystematicerrorof of3is∼25%.InthecaseofAV264avalueof3wasdeemed
0.05dexinlogg.Inadditiontothisthereisarandomerrorof more appropriate from a fit to the redward wing of the emis-
±0.05fortheB2-B2.5starsandupto0.1dexfortheB0-B1.5 sionprofileandtheheightoftheemissionpeak.β parameters
stars(SeeTable2). forAV373& AV151areas uncertainas thatforAV96dueto
Microturbulent velocities, ξ , adopted for the analysis the failuresin thecodeto producecompleteemissionprofiles
Si
were derived from the Si III triplet at 4560 A˚. The microtur- atlowtemperatures(seebelow)andwhichcontributetotheer-
bulencewasalso derivedfromtheO II multiplets,inthestars rorinthemass-lossratederived.Uncertaintiesinthemass-loss
withspectraltypeearlierthanB2.Asdiscussedinpaper1these rates therefore range from 15% to 25% and are presented in
valuestendtobe∼10kms−1 higherthanestimatedfromthe Table2.
silicon lines. In the case of the stars cooler than B2, no O II The final model fits to the observed Hα profiles are pre-
lines are present in the spectra as a result of the low temper- sentedinFigs.1.Threeoftheobjects,AV78,AV443&AV56
aturesand hence low ionisation of this species. For AV56the display strong broadening in their profile wings (similar to
spread in equivalent widths amongst the Si III multiplet was AV362&AV22inpaper1).Thesewings,whichareformedin
only∼ 50mA˚ andthiswasnotsufficientto providean accu- the upperlayersofthe photosphere,areconsiderablyaffected
ratemicroturbulentvelocity(typicallythisspreadis>100mA˚ byincoherentelectronscattering.Theintroductionofelectron
intherestofthesample).InadditionAV56istoocoolforitto scattering into the formal solution of the FASTWIND models
haveawelldevelopedOIIspectrum,preventinganestimation providesanexcellentprofilefittoAV78(B1Ia+;seeFig.1).
4 C.Trundle&D.J.Lennon:SMCB-typeSupergiantsII
Fig.1. HαprofilesforthetenSMCB-typesupergiants.Thesolid-dashedlines(---)arethebestfitsfromtheFASTWINDmodels
and the parameters are given in Table 2. In the case of AV78, AV56 & AV443, models with the best fit parameters but with
incoherentelectronscatteringincludedintheformalsolutionarealsoshown(dottedline;···).
Table 2. DerivedatmosphericandwindparametersforSMC B-typesupergiants.Themicroturbulencefromboththe Si III & O II linesare presented,howeverthesilicon
mircoturbulencewasadoptedinthe analysis. A microturbulenceof 10kms−1 wasadoptedforAV56dueto lack ofa directmethodof measurement.Escape velocitiesare
calculatedusingtheprocedureoutlinedinpaper1&EVANS04.Terminalvelocities,(v ),forAV242,AV264&AV78arederivedfromtheSEImethod(1).Fortherestof
∞
thesample,v∞iscalculatedbyadoptinga vve∞sc ratioof2.65forstarswithTeff ≥21.0kKand1.40forTeff <21.0kK(2).TheerrorsquotedarediscussedinSection3.
Star Teff logg R⋆ Mspec Mevol log(LL⋆) ξSi ξO M˙ v∞ vesc β log(DMOM)
⊙
(kK) (cgs) (R⊙) (M⊙) (M⊙) (kms−1) (10−6M⊙yr−1) (kms−1) (kms−1) (cgs)
AV420 27.0±1.5 3.05±0.15 21.7 19 26 5.35 13 25 0.34±0.15 1310±2602 4932 1.0 28.11±0.40
AV242 25.0±1.5 2.85±0.15 36.6 35 39.5 5.67 13 20 0.84±0.13 950±1001 3591 2.0 28.48±0.15
AV264 22.5±1.5 2.55±0.15 34.8 16 29 5.44 13 19 0.29±0.06 600±1001 3131 2.5 27.81±0.20
AV78 21.5±1.5 2.40±0.15 79.0 57 53 5.92 12 20 2.29±0.34 450± 501 4231 3.0 28.76±0.15
AV96 22.0±1.5 2.55±0.15 34.0 15 27 5.39 11 21 0.24±0.06 850±1702 3202 3.0 27.87±0.25
AV373 19.0±2.0 2.30±0.20 46.8 16 28 5.42 11 22 0.16±0.04 390± 802 2822 3.0 27.42±0.25
C
AV10 17.0±2.0 2.20±0.20 46.7 13 22 5.21 14 0.15±0.03 380± 752 2732 3.0 27.39±0.20 .T
r
AV56 16.5±2.0 2.05±0.20 96.1 38 50 5.88 10 0.51±0.08 420± 852 3022 2.0 28.12±0.15 un
AV443 16.5±2.0 1.95±0.20 96.5 30 42 5.79 11 0.45±0.09 340± 702 2462 2.0 27.97±0.20 dle
AV151 16.0±1.5 2.10±0.15 57.1 15 24 5.28 15 0.16±0.04 370± 852 2632 3.0 27.45±0.25 &
D
.J
.
L
e
n
n
o
n
:
S
M
C
B
-
ty
p
e
S
u
p
e
rg
ia
n
ts
I
I
5
6 C.Trundle&D.J.Lennon:SMCB-typeSupergiantsII
Howeveritdoesn’tcompletelyaccountforthestrongwingsob- ing uncertainties in these abundances are the adopted micro-
servedinthetwocoolerB2.5Iatypestars;thiswasalsothecase turbulence,ξSi,equivalentwidthsandatomicdata.Inaddition
forAV362(B3Ia)&AV22(B5Ia).Alsonotethemodelsfailure the lower resolution of the EMMI data causes a slight relative
toreproducetheobservedprofilesforspectraltypeslaterthan increase in the uncertainty of the equivalent width measure-
B2Ia(typicallyforT below19kK).Themodelsbelowthese ments, particularly for closely spaced, weak lines which may
eff
temperaturesexhibitstrong absorptionfeatures not present in havebeenresolvedintheUVESdatabutareslightlyblendedin
theobservedspectra.Inthesemodels,theLymancontinuumis theEMMIdata(viz.SiIV4116A˚ &HeI4121A˚).
opticallythickcausingthetransitionfromthe1stto2ndenergy One difference in this work and that carried out for the
levelstobeindetailedbalanceandthesecondenergylevelthen UVES data are the number of nitrogen and oxygen lines con-
simulatesa groundstate. TheHα line (i.e.transitionfromthe sideredintheabundanceanalysis.ExperiencewiththeB-type
2nd to 3rd energylevel)willsubsequentlyresultin a P-Cygni supergiantspectra identified some additional, reliable lines to
profile as it behaves like a quasi-resonanceline. This may, in be includedin the analysis(viz. O II 4069, 4349& 4676and
part, be due to the current temperature structures used in the N II 4241A˚). In generalthe weakestlines in the UVES abun-
FASTWIND models and could possibly be improved with the danceanalysiswerenotmeasurableintheEMMIdataandwere
introduction of a temperature correction scheme. A new ver- omittedfromtheanalysis.
sion of FASTWIND has recentlybeen released which amongst Following the procedures laid out in Paper 1, the carbon
otheradvancesincorporatessucha scheme (Pulset al.2005). abundancewascorrectedforNLTEeffectsduetotheproblem-
ItisbasedonanalternativemethodintroducedbyKuba´t,Puls atic line 4267 A˚. Table 3 states a mean carbon abundance of
& Pauldrach (1999) to solve the radiative equilibrium equa- 6.89dex,byapplyingthe appropriatecorrectionof+0.34dex
tionexplicitly.Withthistechnique,thetemperatureintheupper ameanCabundanceof7.23dexisestimatedforthissample.
photosphereandwindiscalculatedbyensuringthatathermal Itisworthnoting,thatpreliminaryabundanceestimatesfor
balanceofelectronsisobtained. 64SMCB-typesupergiantsweremadebyDuftonetal.(2000)
OnenoticeablefactorfromTable2isthelargevaluesofthe in a studyofline strengths(thisworkused thecompleteatlas
β-parameterdeterminedfortheseB-typeobjects.Thesediffer ofLennon1997).TheMgandSiresultsareincloseagreement
fromthoseobservedandpredictedforO-typestarswhichhave with those foundhere, whilst the C, N andO abundances,al-
much faster winds with β being close to the order of unity. though lower differ by no more than 0.21 dex (see Table 4).
Indeed large β values have been observed in other optically Duftonetal.founda rangeofnitrogenabundancesfrom7.09
basedspectroscopicstudiesofB-typesupergiantsintheGalaxy to7.69dextofitthedata,whilstthemaximumvalueissmaller
and SMC (Kudritzki et al. 1999; Evans et al. 2004a). On the thanthatderivedfromthemoredetailedanalysisinthiswork
otherhandapplyingtheSEI-methodtoUVP-Cygnilinesofthe there is some consistency between the spread in results from
SMCtargetspresentedhereandinPaper1andoftheGalactic the two analyses. This work shows that with a large sample,
targetsanalysedinKudritzkietal.suggestlowervaluesofβof even basic analysis techniques can arrive at reasonable esti-
1-1.5(EVANS04andHaser1995,respectively).Inboththese matesforthechemicalcompositionsofastellarpopulation.
UV analyses it was suggested that whilst small changes of β
can give aesthetically more pleasing fits to the UV resonance
5. Discussion
lines,theywererelativelyinsensitivetotheparameter.Infact,
Haser (1995) found that the SEI method of the B-type stars Foracomparisonofourderivedabsoluteabundanceswiththe
produce identical profiles for β values in the range 1-4. It is results of previous studies in the SMC, we will assume that
likely that the relatively slow winds of these objects result in the H II regionabundancesofKurtetal. (1999)representthe
theP-Cygnilinesundersamplingthevelocityofthewindover present day composition of the SMC. In addition we include
thesmallrangeindepthforwhichtheformationoftheselines theresultsofRollestonetal.(2003;correctedfornon-LTEef-
cover,incomparisontotheHα profile.Hencetheβ parameter fects as discussed in Paper 1) for the main-sequence B-star,
isilldefinedbytheUVlinesintheseslowerwindobjects. AV304, whose photosphereappears to be uncontaminated by
anybyproductsofnuclearprocessesinthestellarcore.Thus,
it is assumed that the chemical make-upof these objects rep-
4. ChemicalComposition
resent the initial baseline composition of the SMC (for more
Theproceduresfortheabundanceanalysisareidenticaltothat on this topic see paper 1). The mean element abundance for
discussedinPaper1,theyapplythesameatmosphericmodels thissampleispresentedinTable3,butforthepurposeoffur-
andatomicdataandassuchonewouldexpecttheabundances ther discussion the results from all 17 supergiants and 1 gi-
to follow the same patterns. Indeed, no surprises are hidden antwillbeconsidered.HenceTable4presentsthemeanabun-
in the chemical composition of these 10 B-type supergiants. dancescalculatedbycombiningtheresultsinTable3andthose
Consequently,afewinterestingpointswillbeintroducedhere presented in Table 5 of paper 1. The results from SMC A-
with further discussions of the implications in the following type supergiants and other B-type stars are also included in
section. Table 3 (Venn1999;Venn & Przybilla 2003;Lennon,Dufton
Fromagridofequivalentwidthsgeneratedby FASTWIND & Crowley 2003, these are corrected for non-LTEeffects see
models at various microturbulent velocities and abundances, Paper1).
the absolute abundances of C, N, O, Mg & Si for the SMC AcloseexaminationofTable4revealsthattheα-processed
starswerederivedandarepresentedinTable3.Thedominat- elements(O, Mg, & Si) are in agreementfor all objects,with
C.Trundle&D.J.Lennon:SMCB-typeSupergiantsII 7
Table3. Derivednon-LTEabsoluteabundancesforSMCB-typesupergiants.Abundancesaregivenaslog[N(X)/N(H)]+12,
where X represents the element under consideration.The errors representthe standard deviation of the mean and accountfor
systematicerrors.Thenumbersintheparenthesisarethenumberoflinesincludedintheanalysis.Meanelementabundancesare
presentedinthefinallineofthetable.
Star CII NII OII MgII SiII SiIII SiIV
AV420 <7.08(1) 7.44±0.14(2) 8.06±0.27(10) 6.84±0.24(1) 6.62±0.12(3) 6.59±0.46(1)
AV242 <6.88(1) 7.40±0.19(1) 8.24±0.27(8) 6.84±0.38(1) 6.87±0.10(3) 6.91±0.30(1)
AV264 <6.61(1) 7.86±0.13(3) 8.01±0.25(13) 6.91±0.14(1) 6.65±0.14(3) 6.64±0.28(1)
AV78 <6.90(1) 8.30±0.24(7) 7.86±0.16(12) 6.95±0.16(1) 7.22±0.11(3) 7.26±0.45(1)
AV96 6.88±0.22(3) 7.67±0.23(8) 8.09±0.22(13) 6.97±0.25(1) 6.85±0.21(3) <6.79:(1)
AV373 6.89±0.19(3) 7.46±0.25(2) 8.37±0.39(8) 6.94±0.30(1) <6.79(2) 6.74±0.15(3)
AV10 6.59±0.14(3) 7.66±0.26(3) 8.13±0.31(3) 6.81±0.12(1) <6.76(2) 6.69±0.16(3)
AV56 7.19±0.53(3) 8.27±0.32(2) 6.68±0.09(1) <6.60(2) 6.64±0.22(3)
AV443 6.99±0.36(3) 7.96±0.22(2) 6.72±0.08(1) <6.55(2) 6.63±0.18(3)
AV151 6.85±0.12(3) 7.55±0.19(2) 6.70±0.08(1) 6.54±0.11(2) 6.61±0.18(3)
Mean 6.89±0.18 7.76±0.28 8.11±0.16 6.84±0.11 6.65±0.12 6.75±0.18 6.84±0.27
Table 4. Comparison of mean abundancesfor the complete SMC B-type supergiantsample (including results from paper 1)
with previous studies of abundances in the SMC. Included are the mean abundances of A-type supergiants from Venn et al.
(1999,2003;Si correctedfor non-LTEeffects.),SMC main-sequenceB-star AV304(Rolleston et al. 2003;correctedfor non-
LTE effects), NGC330 cluster main-sequence B-stars (Lennon, Dufton & Crowley 2003; corrected for non-LTE effects) and
SMCHIIregionsfromKurtetal.(1999).ThenumbersinitalicsrepresenttheabundancespredictedfromEWmeasurementsin
SMCB-typesupergiantsbyDuftonetal.(2000;D00).
Supergiants Bstars HII
Thiswork D00 A-type NGC330 AV304 regions
C 7.27±0.14 7.10 7.26±0.15 7.41±0.18 7.53±0.06
N 7.71±0.32 7.49 7.52±0.10 7.51±0.18 6.55±0.01 6.59±0.08
O 8.13±0.13 7.95 8.14±0.06 7.98±0.13 8.16±0.33 8.05±0.05
Mg 6.81±0.14 6.78 6.83±0.08 6.59±0.14 6.73
Si 6.75±0.18 6.78 6.92±0.15 6.58±0.32 6.74±0.03 6.70±0.20
the possible exception of the NGC330 stars. The abundances results from the present study, the plot includes the abun-
oftheα-processedelementsintheNGC330starsappeartobe dancesderivedfor OBA-typestars by Venn(1999),Crowther
lowerthantheotherSMCobjectsby≤0.25dex.Thiscluster etal.(2002),Lennon,Dufton,&Crowley(2003),Hillieretal.
was found to be 0.5 dex metal deficient with respect to SMC (2003), Venn & Przybilla (2003)& Bouret et al. (2003). The
field stars in a photometric study made by Grebel & Richtler histogram clearly illustrates that there is a large range in ni-
(1992).Howeverthe resultshere concurwith the lowermetal trogen abundances observed in these stars. The majority of
deficiency of this cluster observed in recent analyses by Hill the objects have enhanced abundances, which predominantly
(1999)andLennon,Dufton&Crowley(2004)onKandB-type lie around∼7.50dex,closertothebaselinegalacticnitrogen
stars,respectively.CNprocessedmaterialisonceagainevident abundance than that of the SMC. The stars considered have
intheatmospheresoftheB-typestars,thenitrogenabundances stellar masses inthe range11to 60M , thustodecipherany
⊙
in the overallsample studiedherevariesfrom7.14dexin the correlationwith mass andsubsequentlyluminositythe results
giant, AV216, to 8.30 dex in the most luminousobject in the for the 25 M (box outlinedin bold dashed line) and 40 M
⊙ ⊙
sample,AV78.Thelowestnitrogenenhancementinthesuper- (shadedbox)OBstarshavebeenoverplottedinFig2(noabun-
giantsampleisafactorof7(7.40dex)abovethenitrogenabun- dancesforAstarsinthismassrangeareavailableinthelitera-
danceofAV304andthenebularresults.Thisenrichmentisin ture).Whilstthereappearstobeatrendofnitrogenabundance
close agreement with the results from the A-type supergiants withmass,inthathighernitrogenabundancesareproducedin
andB-typemain-sequencestars(Venn1999;Venn&Przybilla themostmassivestars,itisunwisetomakesuchdefinitivecon-
2003;Lennon,Dufton,&Crowley2003). clusionsfromthepoorstatisticsavailabletousatpresent.
To understand the distribution of nitrogen abundances in Recallthatthesignatureofphotosphericcontaminationby
OBA-typestarsintheSMC,ahistogramwasassembledfrom CN-cycled material is an enhancementin nitrogenwith some
the results in the literature (see Fig 2). In addition to the depletion of carbon and possibly oxygen.No obvioussign of
8 C.Trundle&D.J.Lennon:SMCB-typeSupergiantsII
Fig.3. HR-diagram,luminosityas a functionof temperature.
The stellar evolution tracks of Maeder & Meynet (2001) are
shownforanassumedinitialrotationalvelocityof0(—;solid
lines)and300kms−1 (--;dashedlines)atsolarmassesof15,
20,25,40,&60M .Includedaretheresultsfromtheearlyand
⊙
midB-type supergiantsof boththe UVES (◦& openstar) and
EMMI (• & ⋆) data, with errors in luminosity representing ±
15%.Inaddition,theSMCO-typesupergiantsofHillieretal.
Fig.2. HistogramofresultsfromabundanceanalysesofSMC (△;2003)andCrowtheretal.(⋄;2002),SMCO-typedwarfs
OBA-type stars collectedfrom the literature.The dotted lines
ofBouretetal.(opensquare;2003)andAF-typesupergiants
represent the baseline nitrogen abundancesin the SMC (6.55
fromVennetal.(+;1999,updatedby2003)arealsoshown.
dex)andMilkyWay(7.70dex).Thetoppanelshowsthenitro-
genabundanceofO-typestarsintheSMCusingtheresultsof
supergiantsofCrowtheretal.(2002)andHillier etal.(2003) hasahighercarbonabundancethantherestofthesample,nor
and SMC O-type dwarfs of Bouret et al. (2003). The mid- thatAV215andAV78havedepletedcarbon.
dlepanelillustratesthedistributionofobservednitrogenabun-
dancesfromB-starsintheSMC,theseincludetheresultsfrom
5.1. EvolutionofB-type Supergiants
this work and the B-type stars of Lennon,Dufton& Crowley
(2003).Thebottompanelincludesthenitrogenabundancere- Inpaper1itwasshownthatstellarevolutionmodelswhichin-
sults of the A-type supergiantsin the SMC from the analysis clude the effects of rotation can reproduce,for the most part,
of Venn et al. (1999) with the updated values from Venn & the observed nitrogen abundances in the SMC B-type stars.
Przybilla(2003).Inthetopandmiddlepanelthedistributionof For the purpose of this discussion, the complete sample will
starswithMevol ∼25M⊙ (boxoutlinedinbolddashedlines) againbecomparedtotheevolutiontracksofMaeder&Meynet
and 40 M⊙ (shaded box) from the original set of targets are (2001)whichhavebeenre-calibratedtoametallicityof0.2Z⊙
alsoindicated. (see Paper 1) and assume an initial rotationalvelocity of 300
kms−1.ThepositionofthisstellarsampleontheHRdiagram
isillustratedinFig3.Fromthesetracksallofthestarsseemto
depletionsin oxygenhave been detected in this sample of 18 bepasttheendofthehydrogenburningmain-sequencephase,
stars,consistentwiththeoreticalpredictionsthatindicatemod- withsomeoftheobjectshavingthickhydrogenburningshells.
eratedepletionsinoxygen,smallerthantheerrorsinouranal- Indeed the loci of the cooler stars (B2.5 - B5) on the evolu-
ysis.ItisinterestingtonotethatAV215andAV78(whichhave tionary tracks imply that they are in the core helium-burning
the highest N enhancements of 26 and 56 times the baseline phase.
value, respectively), have the lowest oxygen abundances, but Inthesecomparisonsonemustbecarefulastheactualstars
thesevaluesarestillwithintheerroronthemean.Inaddition, mayhavearangeofinitialrotationalvelocitiesratherthanthe
the mean carbon abundance of the B-type supergiant sample 300kms−1assumedin the evolutionarymodels,andtherefore
and the NGC330 stars, are lower than that of AV304 and the the tracks presented in Fig 3 may be inappropriate for some,
HII regions,whetherthisisasignificantdifferenceisdifficult orall,ofthestars.Includingrotationalvelocitiesintheevolu-
toascertaingiventheuncertaintiesintheabsoluteabundances tionary modelshas the effectof extendingthe main-sequence
ofthision.WithintheB-typesupergiantsamplethereisnoev- branchtohigherluminositiesandtolowertemperatures,which
idence that AV216, (the giant and hence the least processed), willthereforeresultinlowerevolutionarymassesforastarofa
C.Trundle&D.J.Lennon:SMCB-typeSupergiantsII 9
Fig.4. Comparisonofevolutionaryandspectroscopicmasses Fig.5. Surfacenitrogenabundancesasafunctionoftempera-
as a function of luminosity. Included are the results from the ture.The stellar evolutiontracksofMaeder& Meynet(2001)
earlyandmidB-typesupergiantsofboththeUVES (◦&open are shown for an assumed initial rotational velocity of 300
star) and EMMI (• & ⋆) data. Additionally the results from kms−1 andvarioussolarmasses.Includedaretheresultsfrom
NGC330 B-type stars from Lennon, Dufton & Crowley (∗; the early and mid B-type supergiants of both the UVES (◦ &
2003)areshown.Theerrorbarsrepresenttheerrorinthesur- open star) and EMMI (• & ⋆) data. Also displayedare the re-
facegravity.Also shownis theuncertaintyin massduetothe sultsfromSMCO-typesupergiantsofHillieretal.(△;2003)
adopteddistancemodulus(--;dashedlines). andCrowtheretal.(⋄;2002),SMCO-typedwarfsofBouret
et al. (open square; 2003), NGC330 B-type stars of Lennon,
Dufton, & Crowley (∗ ; 2003), and A-type supergiants from
givenluminosity.Starswithinitialstellarmasses>15M⊙tend Vennetal.(+;1999,updatedby2003).Theerrorbarsillustrate
to cross the HR-diagram from the end of the main-sequence therandomandsystematicerrorsonthenitrogenabundance.
phasetowardscoolertemperatureswithrelativelyconstantlu-
minosities, on the way to becoming red supergiants (RSG).
Maeder& Meynet(2001)showeda comparisonof evolution-
ary tracks adopting initial rotational velocities of 0, 200, 300 some,ifnotall,oftheoverestimationinMevol.Anotherpossi-
and 400 kms−1, which implies only a small difference in lu- ble explanationis that the lower luminosity stars may be in a
minosity (∆log(L⊙/L⋆) < 0.1) as the star movesacross the post-RSG blue-loopphase and thatthe lifetime for this phase
HR-diagramforaninitialstellarmassofM⋆=20M⊙(seeFig. isgreaterthanthepre-RSGstage.
8inMaeder&Meynet2001).Itisthereforereasonabletoes- Toappreciatethesuccessofthenewevolutionarymodels,
timate evolutionarymasses fromthese tracks, with the caveat whichconsiderrotation,inreproducingtheobservednitrogen
that the assumed initial rotational velocities and the errors in abundances in hot, luminous stars it is useful to look at the
interpolating between the tracks lead to uncertainties of ∼ 5 productionofnitrogenasafunctionoftemperatureforvarious
solar masses (and possibly more at the high end of the mass stellar masses (see Fig. 5). The models with stellar masses in
rangeunderconsideration). therange20-60M havenitrogenenrichmentsbytheendof
⊙
The estimated evolutionary masses, M , based on the thecorehydrogenburningphaseofafactorof7to20abovethe
evol
300kms−1tracksaredisplayedinTable2alongwiththespec- baselinenitrogenabundanceof6.55dex(fromAV304).Thisis
troscopicmasses,M . Theseexhibitthesame behaviouras similartotheenhancementsobservedinoursample,although
spec
found from the analysis of the UVES data, such that Mevol > someofthemoreluminousobjectshavegreaterenhancements.
M by approximately0.10 - 0.30 dex (see Fig. 4). The er- Indeed, adopting a stellar mass for each object from the HR-
spec
rors in the adopted distance and gravity account for much of diagraminFig3andfollowingtheappropriateevolutiontrack
thediscrepancy,leavingonlya differenceof0.10dex(or∼8 inFig.5,itwouldappearthatmanyoftheobjectshavegreater
M ). It is difficultto accountfor the remainingdisagreement enhancementsthanpredicted.Howevergiventheerrorsinthe
⊙
unlessitisanoverestimationofM duetoerrorsininterpo- abundancedeterminationsthismaynotbesignificant.Thetwo
evol
lation. However, the largest discrepancies are observedin the B2 Ia stars (AV373 & AV18) are interesting objects as they
less luminous,and henceless massive stars. Inthese stars the haveverysimilaratmosphericandwindparameters,inaddition
tracksaremorewidelyspacedandthereforemoreaccuratees- theirnitrogenabundancesarethesame,withintheerrors.This
timatesofM wouldbeexpected.Byincludingsomedegree is also the case for the two B0.5 Ia stars, AV420and AV104.
evol
ofconvectiveovershootingthetrackforagivenmasswouldbe IndeedthepositionofthesefourstarsontheHR-diagramfall
extendedtohigherluminositiesandcouldaccountforatleast close to the theoreticalpredictionsfora 25M star and have
⊙
10 C.Trundle&D.J.Lennon:SMCB-typeSupergiantsII
abundancesinagreementwiththemodelsfora25M starwith
⊙
initialrotationalvelocitiesof300kms−1.
Fromlinedrivenwindtheory,itispredictedthatthestellar
winds are stronger in stars of higher metallicity (i.e. that the
mass-loss rates are higher). This comes directly from the de-
pendenceofthephotonmomentumtransfertothewindonthe
number of optically thick lines and that at lower metallicities
thelinestrengthsdecrease.Ifthisisthecase,thelossofangular
momentumbytheendofthemain-sequenceismuchgreaterin
galacticobjectsthaninlowermetallicitystars(viz.SMC).This
reductioninangularmomentumatforexample,solarmetallic-
ity,leadstoamorerapiddecreaseinstellarrotationalvelocity.
Therefore, the models by Maeder & Meynet (2001) for stars
withZ=0.004(SMCmetallicity)havehigherrotationalveloc-
ities after core hydrogenburning than similar stars with solar
metallicity. Since these stars have higher rotational velocities
theconvectivezonesarelargerandtheatmosphericdistortions
moresevere,thisresultsin greaterenhancementsin thenitro-
gencontentofthephotosphere. Fig.6. WLR as derived from SMC B-type supergiants. The
Do we observe this variation in rotational velocities from plotshowstheearly(circles)andmid(stars)B-typestarsfrom
evolvedstarsinthetwodifferentenvironments?Unfortunately theSMCsample(UVES:opensymbols;EMMI:filledsymbols).
thisquestionisdifficulttoanswerbasedonspectraalone.We Included are the linear regression fits to the observationalre-
doseegreaternitrogenenrichmentsintheSMCstars,yetitis sultsofthe early(all)andmid (only<B3) B-typestars (—-;
not possible to ascribe this to rotational velocities at present. solid lines). Thedashedlines (- - -) arethe wind-momentum-
vsini values have been measured for OBA-type stars in the luminosityrelationship,fromgalacticB-typesupergiantswith-
galaxyandSMCbuttheestimatesforbothregionsaresimilar outline-blanketing(Kudritzkietal.1999).Theerrorbarsrep-
(intherange40-150kms−1). Howeverasnotedpreviouslyin resenttheuncertaintyinderivingthewindmomentum.
Sect.2,thesemeasurementsarenottrulyrepresentativeofthe
projectedrotationalvelocitybutaredominatedbymacroturbu-
stars, it is more appropriate to make the early/mid B distinc-
lence.Also note thatthese are theprojectedrotationalveloci-
tionbasedonstellartemperaturesratherthanspectraltype.The
ties and some information on the inclination of the rotational
hottest B1.5 star in this sample (Sk191)has an effective tem-
axis is requiredto enable the true velocity to be disentangled
peratureof22.5kK.InPaper1itwasshownthatthisobjecthas
fromvsinimeasurements.
awindwhichbehavessimilartothecoolerstars,thereforethis
has been adopted as the cut off temperature for early B-type
5.2. Mass-loss: ObservationsandPredictions. starsinthefollowingdiscussion.ThisimpliesthatAV264and
AV78 may exhibit signs of having lower wind momenta than
In this section the results of the entire sample of B-type su- thehotterB1stars,butsimilartothatofthemidB-typestars.
pergiants will be discussed in an attempt to glean more in- Adirectcomparisonofthewindmomentaforthesestarswith
formation on the behaviour of their stellar winds as a func- thoseofthehotterB1-typeobjectsisnotappropriateduetothe
tionofluminosity,spectraltypeandmetallicity.Fromasimilar luminositydependenceofthestellarwind.
analysisof B-type stars in the Galaxy,Kudritzkiet al. (1999) No clear distinction between the wind-momenta of the
found a clear distinction between the wind momenta in early SMC sample and the WLR of their galactic counterparts is
andmidB-typestars.Thisbehaviouralsodifferedfromthatof evident. This consistency still holds if one regards the wind-
theO-typestarsfromPulsetal.(1996).Thischangeinwind- momenta of the individual Galactic and SMC B-type super-
momentawithspectraltypeisdependentonthefluxweighted- giants at a certain luminosity, although a larger scatter is ob-
distributionofthespectrallines,theionisationofthelinescon- servedintheSMCmidB-typestars.Itseemsappropriatetoat-
tributing to the line force and the relative number of strong temptanestimationoftheWLRfromtheseSMCtargetssince
toweak linesdrivingthe wind.Thechangein wind-momenta theyare the firsthomogeneousresultsfromB-typesstarsin a
observedin B-typestarsbyKudritzkietal. occurredbetween low metallicity environment. We adopt the same form of the
twoobjectswithspectraltypesB1andB1.5atTeff=23.5and WLRasdefinedbyKudritzkietal.(1999)
22.5 kK (note these temperatures were determined from un-
blanketed models). This is approximately the point in which logDmom =logDo+xlog(L⋆/L⊙) (1)
the so-called ’bistability jump’, a change in the properties of
thewind,ispredictedtotakeplace(Lamers,Snow&Lindholm wheretheinverseoftheslope,1/x=α′.α′describesthedepth
1995). dependenceoftheradiativelineforceandcanbeexpressedas
DuetotherangeintemperaturesobservedfortheB1stars a function of the force multiplier parameters of the radiative
inthissampleandthefactthatitoverlapswiththatoftheB1.5 lineforce(i.eα′ =α−δ).D isaproductofthemass-loss
mom